Molecular and biophysical features of hippocampal “lipid rafts aging” are modified by dietary n‐3 long‐chain polyunsaturated fatty acids

Abstract “Lipid raft aging” in nerve cells represents an early event in the development of aging‐related neurodegenerative diseases, such as Alzheimer's disease. Lipid rafts are key elements in synaptic plasticity, and their modification with aging alters interactions and distribution of signaling molecules, such as glutamate receptors and ion channels involved in memory formation, eventually leading to cognitive decline. In the present study, we have analyzed, in vivo, the effects of dietary supplementation of n‐3 LCPUFA on the lipid structure, membrane microviscosity, domain organization, and partitioning of ionotropic and metabotropic glutamate receptors in hippocampal lipid raffs in female mice. The results revealed several lipid signatures of “lipid rafts aging” in old mice fed control diets, consisting in depletion of n‐3 LCPUFA, membrane unsaturation, along with increased levels of saturates, plasmalogens, and sterol esters, as well as altered lipid relevant indexes. These changes were paralleled by increased microviscosity and changes in the raft/non‐raft (R/NR) distribution of AMPA‐R and mGluR5. Administration of the n‐3 LCPUFA diet caused the partial reversion of fatty acid alterations found in aged mice and returned membrane microviscosity to values found in young animals. Paralleling these findings, lipid rafts accumulated mGluR5, NMDA‐R, and ASIC2, and increased their R/NR proportions, which collectively indicate changes in synaptic plasticity. Unexpectedly, this diet also modified the lipidome and dimension of lipid rafts, as well as the domain redistribution of glutamate receptors and acid‐sensing ion channels involved in hippocampal synaptic plasticity, likely modulating functionality of lipid rafts in memory formation and reluctance to age‐associated cognitive decline.

Unique in the brain tissue, it is the fact that nearly all lipids are contained in plasma and intracellular membranes, which, in turn, are organized in highly dynamic microdomains. A particular type of membrane microdomains named lipid rafts has received considerable attention. Lipid rafts are liquid-ordered domains enriched in membrane lipids such as cholesterol, sphingolipids, saturated acyl chains, and gangliosides (GM1), which create a physicochemical environment for molecular interactions favorable for the compartmentalization of protein clusters governing signaling pathways (Egawa et al., 2016;Lorent & Levental, 2015;Suzuki et al., 2011). The molecular clusters are altered in response to pathophysiological conditions, such as oxidative stress and neurodegeneration (Allen et al., 2007;Díaz & Marin, 2021;Marin & Diaz, 2018).

Recent studies have shown that NMDA receptors (NMDARs)
activation in the hippocampus involves translocation to synaptic lipid rafts during plasticity processes necessary for spatial memory formation (Bannerman et al., 2014). Spatial training induces a rapid and selective translocation of NR1 and NR2A NMDA-R subunits and scaffolding PSD-95 to synaptic lipid rafts and decreases the soluble fraction of postsynaptic membrane (non-raft PSD) (Delint-Ramirez et al., 2010). Furthermore, lateral reallocation of AMPA receptors affecting synaptic strength has been demonstrated between synaptic nanodomains under experimental short-and long-term potentiation (STP and LTP) conditions (Groc & Choquet, 2020).
LCPUFA content has generally been observed to drop (unevenly) during aging in different brain areas. This may be caused by altered fatty acid metabolism (including membrane phospholipid biogenesis), lower rate of transport of LCPUFAs through the blood-brain barrier, and peroxidation of LCPUFA pools (Catalá & Díaz, 2016;Fabelo et al., 2014;Naudí et al., 2015;Yoon et al., 2022). A wealth of studies on aging-associated changes in brain lipids indicate that besides LCPUFA depletion, other important lipid components of nerve cell membranes are also altered in an age-and region-specific manner (Colin et al., 2016;Ledesma et al., 2012;Naudí et al., 2015).
Although much less explored, brain lipid rafts undergo complex lipid modifications associated with aging, which likely impact synaptic transmission. We have previously described a complex set of changes in the contents of different fatty acids, lipid classes, and biophysical properties in brain cortex lipid rafts from male mice associated with aging (Diaz et al., 2012). This phenomenon, termed "lipid raft aging", was subsequently demonstrated in lipid rafts from the human frontal cortex (Díaz et al., 2018;Fabelo et al., 2014). Lipid raft aging is characterized by changes in the lipid structure of the raft bilayer are caused by a marked reduction in membrane unsaturation and peroxydability and cholesterol-to-phospholipids ratio, increased sphingolipids (sphingomyelin and sulfatides), saturated fatty acids, and sterol esters , which collectively render lipid rafts more viscous and less fluid (Diaz et al., 2012;Díaz et al., 2016).
There exists considerable interest in the possible role of dietary fatty acids in age-related cognitive decline and cognitive impairment related to Alzheimer's disease, raising the possibility that dietary LCPUFA may be incorporated as nutraceuticals for the prevention of age-associated neuropathologies. Indeed, several observational and epidemiological studies have suggested that saturated fatty acids have negative effects on cognitive function and that dietary intakes of PUFAs are associated with reduced risk for age-associated cognitive decline and protection against the risk of AD (Solfrizzi et al., 2018;Zhang et al., 2020).
Moreover, experimental evidence has shown that nerve cell membrane LCPUFA may be modified by dietary interventions Yaqoob & Shaikh, 2010). However, one common observation from nutritional studies is that both, diets deprived of, or enriched in, LCPUFA or their precursors, have a rather discrete (yet significant) effect on the overall brain lipid composition, despite severe nutritional interventions.
This clearly limits the efficiency of dietary manipulations to correct ageassociated depletion in essential fatty acids and related lipid classes but also demonstrates the existence of highly efficient lipostatic mechanisms preserving nerve cell membrane functionality. Nevertheless, the extent to which LCPUFA content affects the neuronal membrane, remains largely unexplored, especially in the context of synaptic plasticity and memory formation. This is particularly relevant in females, where the presence of ovarian hormones are modulatory factor for hepatic and brain lipid metabolism, and because nerve cells lipidome is subjected to a complex interplay between estrogens and endogenous biosynthesis and/or dietary LCPUFA availability .
In the present study, we sought to investigate changes in lipid profiles, physicochemical properties, and distribution of synaptic receptors in hippocampal lipid rafts during aging. We have used cycling female mice.
Estrogens are known to modulate the effects of dietary LCPUFAs on the K E Y W O R D S acid-sensing ion channels, Aging, AMPA-R, arachidonic acid, cholesterol, DHA, dietary LCPUFA, fluidity, ganglioside GM1, lipid rafts, membrane microdomains, mGlu-R, microviscosity, NMDA-R, sphingolipids cerebral cortex  and estrogen decline is associated with a higher incidence of neurological disorders (Marin & Diaz, 2018). We have undertaken a heuristic approach based on global lipid profiles. This allowed us to assess the relevance of lipid variations associated with normal aging in hippocampal lipid rafts. Also, we have investigated the compensatory effects of n-3 LCPUFA supplementation to counteract the effects of aging on lipid raft structure and functionality. The outcomes have disclosed ageindependent modifications on the lipid signature, interdomain remodeling, biophysical properties, and glutamate receptor distribution in female hippocampal lipid rafts in response to n-3 LCPUFA supplementation.

| Aging modifies the lipid composition of Hippocampal lipid rafts
The composition of lipid classes and fatty acids extracted from hippocampal lipid raft fractions in the four experimental conditions are shown in Tables 1 and 2. Lipid profile in 6 months control animals (CTRL6M) fed a standard diet display a close agreement with lipid classes and fatty acid data previously reported for hippocampal lipid rafts under control conditions (Canerina-Amaro et al., 2019;Diaz et al., 2012;Fabelo et al., 2012).
Aging caused a significant change in the lipid profiles of hippocampal lipid rafts under a standard diet. Thus, in 15 months-old animals a fed control diet (CTRL15M), a significant depletion of DHA, AA, LCPUFA, total n-9, as well as reduced UI and PI were observed (Table 1). The relevant UI/sat and PI/sat indexes were significantly reduced. These changes were accompanied by reduced cholesterol and sphingomyelin levels, and cholesterol-to-phospholipids ratio (CHO/TPL), and increased sterol esters (SE), SE/CHO, and n-6/n-3 ratios, as compared to young littermates ( Table 2). The fact that the n-6/n-3 increases with aging may favor a proinflammatory state in the brain parenchyma (Stables & Gilroy, 2011). Similar results have been shown in hippocampal extracts from aged WT and transgenic APP/PS1 mice (Taoro-González et al., 2022).
Another differential feature of PLs compared to PE was their opposed affinity for main LCPUFA, with PlsEtn(a) and PlsEtn(b) associating positively to DHA and negatively to AA, while PE was unrelated to either LCPUFA. These observations are relevant from the point of view of aging since PE and PlsEtn(a) increase, while PlsEtn(b) reduces, in older animals, therefore favoring n-3 depletion of lipid rafts under conditions where these fatty acids supply is limited, such is the standard diet.
Depletion of plasmalogens has been reported as one of the earliest events in the development of AD (Han, 2005). These results are also relevant from the oxidative stress point of view, as plasmalogens have been proposed to behave as LCPUFA depots, to participate in membrane fusion, and also to act as molecular sinks for ROS species (Brites et al., 2004;Skowronska-Krawczyk & Budin, 2020).
Hence, our present results confirm previous studies showing that brain aging is accompanied by a reduction of lipid rafts overall unsaturation and peroxydability, as well as related physicochemical properties (i.e. fluidity), in the context of the so-called "lipid raft aging" (Diaz et al., 2012;Fabelo et al., 2012Fabelo et al., , 2014. In line with this, age-related changes in both total lipid abundance and region-dependent composition have long been observed, with PUFA contents generally dropping during aging (Ledesma et al., 2012;Naudí et al., 2015). However, recent lipidomic studies in whole brain tissue from distinct areas have shown that unsaturation and peroxydability variables either remain stable or may even increase with aging (Mota-Martorell et al., 2022;Naudí et al., 2015). These paradoxical effects of aging can be explained in terms of region-specific membrane lipid metabolism, as well as the differential proportions of rafts and nonrafts domains in bulk brain membranes.

| Dietary LCPUFA supplementation modifies the lipid composition of Hippocampal lipid rafts of young and aged mice
The effects of discrete amounts of LCPUFA in the supplemented diet on the lipid composition of hippocampal lipid rafts are shown in Tables 1 and 2. The most obvious effects are the virtual absence of differences in fatty acids from polar lipids and lipid classes between 6 months (LCPUFA6M) and 15-month-old animals (LCPUFA15M).
Noticeably, raft fatty acid profiles from mice fed the LCPUFAcontaining diet were quite similar to that of 6 months animals fed a standard diet. Indeed, LCPUFA supplementation restored not only DHA contents in lipid rafts, but also levels of saturates (which tended to increase in old standard animals), n-9 monoenes, total n-3, and total PUFA, as well as UI, PI, and their ratios to saturates (PISAT and UISAT, respectively). PE, PlsEtn(a), PlsEtn(b), TPE, AN/ZW, and SE/CHO, were also restored in 15-month-old mice fed the LCPUFA diet. Overall, these observations reflect the regulatory potential of DHA in setting the physiochemical properties of lipid rafts.
Although older animals fed with LCPUFA-enriched diets partly recover the fatty profile of control young animals, significant differences remain in specific fatty acids compared to young animals.
Thus, n-6 LCPUFA arachidonic acid (AA, 20:4n-6) displays the lowest TA B L E 1 Representative fatty acid composition, indexes and ratios obtained in hippocampal lipid rafts in the different groups. Results from two-way ANOVA for main factors and their interactions on each variable is summarized in the right panel.

NS
Note: Data are expressed as mean ± SD for 5-6 animals per group. Different letters in each row indicate statistical differences at p < 0.05 (bold) or 0.05 < p < 0.1 (italic). n-3, n-6 and n-9: first double-bond series.

TA B L E 1 (Continued)
TA B L E 2 Lipid classes composition, indexes and ratios obtained in hippocampal lipid rafts in the different groups. Results from two-way ANOVA for main factors and their interactions on each lipid variable is summarized in the right panel.  values in LCPUFA groups, in particular in LCPUFA 15M (−18.9%).
Further, levels of the n-6 precursors 18:2n-6 (linoleic acid) diminished in LCPUFA groups by −30.9%. The two consequences of these changes are the reduction of n-6 LCPUFA (−17.5%) and the decrease of the n6/n3 ratio (−27.6%) in response to the diet.
Collectively, these data indicate that LCPUFA supply causes the increase of n-3 LCPUFA and the reduction of n-6 PUFA in hippocampal raft membranes. These findings are physiologically relevant, since the lower proportion of the n-6 series, mainly AA, would reduce the potential generation of proinflammatory mediators in the hippocampus (Stables & Gilroy, 2011). Previously, we have observed that dietary n-3 LCPUFA supply reduced age-associated signs of neuroinflammation (Taoro-González et al., 2022).

| Multivariate analyses, main effects and dietaging interactions
Multivariate assessments of lipid fingerprints in both animal and human membrane preparations are accurate Fabelo et al., 2012Fabelo et al., , 2014.  Table 2.
Remarkably, the CTRL15M group segregated and appeared as a differentiated cluster (Figure 1b). One-way ANOVA indicates that factor loadings for PC1 significantly differ between groups, with LCPUFA-enriched groups behaving as a homogeneous cluster, similar to the CTRL6M, but not CTRL15M, groups ( Figure 1c). In the case of factor score 2, differences were only observed between 6M and 15M control groups.
We next checked for internal consistencies of the four experimental groups by means of Cronbach's alpha (α C ) and Intraclass correlation ( LCPUFA groups, which associate in a higher cluster to CTRL 6M group and are totally segregated from CTRL 15M group. These results are in close agreement with the PCA results described above. Finally, it is noteworthy that hippocampal lipid rafts from 6-months animals fed standard and LCPUFA-enriched diets exhibit nearly identical fatty acid profiles, despite large differences in n-3 PUFA supply (Table S1). However, several metabolic correlates may be inferred from the representation of minor fatty acids in lipid rafts. This is the case of 18:3n-3, 18:4n-3, and 18:2n-6cis (Table 1). These fatty acids are located at the entry point of LCFUFA biosynthesis, and their contents are considerably reduced (or absent in the case of 18:3n-3 under standard diet) in lipid rafts from aged animals. Their consistent reduction in older animals indicates that even when LCPUFA are provided in the diet, biosynthetic pathways from n-3 and n-6 PUFA precursors remain functional as a mechanism to provide additional supply to the brain. Further, the facts that 20:5n-3 (EPA) and 22:5n-3 are present in significant amounts in animals receiving the LCPUFA diet (Table 1), independently of age, suggests an efficient mechanism of n-3 LCPUFA storing in hippocampal membranes. Thus, when diets lack n-3 LCPUFA, such the standard diet used here, long chain n-3 and n-6 fatty acids are synthesized from their n-3 and n-6 precursors, 18:n-3 and 18:2n-6, respectively, while such metabolic processes are apparently halted or reduced to a minimum in response to the n-3 LCPUFA containing diet, favoring instead the transport of the longer chain molecule across the blood-brain barrier. This is physiologically relevant, since nervous tissue, in particular hippocampus, does express the genetic and biochemical biosynthetic machinery for PUFA elongation and desaturation (Alessandri et al., 2011;Díaz et al., 2016).

| Changes in lipid rafts microviscosity and estimated fluidity
Fluidity of membranes determines the ease of protein interactions at the initial stages of signaling pathways, but also for conformational changes required for activation of membrane-buried protein, that We detected significant changes in the microviscosity of lipid rafts both at the membrane plane (TMA-DPH) and the hydrophobic core (DPH) of the lipid bilayer. As seen in Figure 3a, estimated microviscosities were significantly higher in CTRL15M for the two fluorescent probes than in any other groups. No differences were observed between 6M and 15M LCPUFA groups which, in turn, were similar to 6M control mice. Average apparent microviscosities (η app ) determined at the membrane plane and hydrophobic core were very similar in LCPUFA groups (Figure 3b). Conversely, for control animals, η app for TMA-DPH was higher than for DPH at the age of 6 months, Finally, lipid packing controls the lateral pressure profiles of membrane domains by the combination of opposed attractive forces (mostly within the hydrophobic core) and repulsive forces (at the head groups) which are balanced in the bilayer structure. Lateral pressure acts on membrane-buried proteins depending on the lipid composition (Lorent & Levental, 2015). Therefore, this physical parameter not only modulates protein function by affecting the equilibrium in protein conformations but also by favoring or dampening interactions in protein clusters. According to our data, lateral pressure is expected to decrease in hippocampal lipid rafts during control aging as a result of the balance between reduction of UI (attractive forces) and augmented PEs (repulsive forces). At domain scale, lateral forces may also participate in the steady-state stability of segregated membrane domains, including the different types of lipid rafts and non-rafts domains.

| Changes on lipid rafts molecular hallmarks and Dimensional features
We have analyzed the contents of two main hallmark constituents of lipid rafts, namely ganglioside GM1, and flotillin-1, in response to the different treatments. GM1 is known to be particularly enriched in the outer leaflet of neuronal lipid rafts. Indeed, the results shown in Figure 4a reveal the nearly exclusive presence of GM1 in lipid rafts compared to non-rafts, with R/NR (raft-to-non-rafts ratio) values ranging from 10 to 1000. The most evident effect in these experiments was the dramatic increase in R/NR values (500-1000 fold) in n-3 LCPUFA groups compared to control groups, which was totally attributable to lipid rafts enrichment since non-rafts GM1 contents did not change between groups. The n-3 LCPUFA-driven enrichment of GM1 is likely due to higher mobilization of gangliosides to membrane rafts. To our knowledge, this is the first report showing the enormous impact of DHA on ganglioside contents in lipid rafts. It may be tantalized that GM1 incorporation into rafts derives from non-rafts membrane reservoirs. This GM1 increase might not be reflected in the corresponding reduction of non-rafts content due to their much larger proportion as compared to lipid rafts. Alternatively, smaller lipid rafts might cluster together into larger lipid rafts, with proportional increase in total GM1 contents, as a consequence of n-3 LCPUFA-induced changes in physicochemical properties of the lipid bilayer. This later hypothesis is supported by evidence from  for domain segregation is represented by the bulkiness of the hydrophilic headgroups of glycosphingolipids and its potential to establish strong conformational associations in glycolipid-enriched clusters (Grassi et al., 2020). These features make ganglioside GM1 a key factor for the stabilization of DHA-enriched raft domains resulting from n-3 LCPUFA supplementation observed here.
Besides their contribution to phase separation and lateral organization of membrane domains, GSLs, and gangliosides in particular, are involved in driving the compartmentalization of "raftophilic" membrane proteins involved in neurotrophic actions, multimolecular complexes such as signalosomes, cellular transduction proteins, ion channels, and transporters, etc. (Allen et al., 2007). These observations have been confirmed in an ever-increasing number of studies, which collectively have led to consider GM1 gangliosides as key factors in neurotrophic and neuroprotective roles (Chiricozzi et al., 2020;Grassi et al., 2020). In order to delve further into the lipid raft modification in response to n-3 LCPUFA, we also analyzed the contents of flotillin-1, a scaffold protein is known to associate with lipid rafts and to participate in the domain stabilization of raft resident and transient signaling proteins (Marin et al., 2009;Marin & Diaz, 2018). This scaffolding structure stabilizes its location within membrane rafts, and it is therefore considered a canonical lipid raft marker. Our results indicate that n-3 LCPUFA treatment brings about a significant increase in flotillin-1 contents in hippocampal lipid rafts, without changes in non-raft membranes (Figure 4b). Consequently, the R/NR is increased by 75-and 27-times in 6M and 15M LCPUFA groups, respectively, compared to age-matched controls.
Together with the results from GM1, these observations demonstrate that LCPUFA induces the reorganization of membrane domains to increase the total membrane surface corresponding to lipid rafts. As the two markers identify targets located at opposite sides of the membrane, and that the response to LCPUFA enrichment was pretty similar, it is concluded that domain reorganization affects both leaflets of the membrane. Taken together, it may be surmised that LCPUFA enrichment leads to the coalescence of smaller lipid rafts into larger lipid rafts (displaying similar membrane order than smaller rafts, but enriched in GM1 and flotillin-1), in a process thermodynamically driven by the aversion between DHA (more abundant in non-raft domains, particularly after n-3 LCPUFA enrichment) and cholesterol (predominant in lipid rafts), leading to higher membrane entropy. Obviously, experimental demonstration of such segregation is technically unfeasible and out of the scope of this study.
Another physiologically relevant outcome obtained in the present study is the effect of aging on GM1 and flotillin-1 contents in membrane domains in control animals. The results in Figure 4c indicate that GM1 and flotillin-1 (FLOT 1) are present almost exclusively in lipid rafts from young animals. This is translated into high R/NR ratios for both markers in 6M CTRL mice. However, these ratios drop by 25% in 15M CTRL animals, indicating that aging provokes the reduction of membrane GM1 and flotilin-1 levels, therefore affecting not only the stability of lipid rafts but also by undermining ligand recognition by raft receptors and the scaffolding function of rafts to promote the recruitment of signaling molecules and receptors (Colin et al., 2016;Westra et al., 2021;Yaqoob & Shaikh, 2010).

| Outcomes from the mathematical model
The results from the mathematical model indicate that the effects of aging and n-3 LCPUFA supplementation can be largely explained in terms of molecular dynamics. Results in Figure 4c Figure 3c) rather than in the membrane plane (TMA-DPH in Figure 3c). At the membrane core, lipid packing is mostly determined by the formation of Van der Waals forces between acyl chains, which expectedly will be larger the lower the UI/SAT ratio, as shown in Figure 3d. Conversely, the model fails at estimating membrane surface fluidity, a drawback that can be interpreted as it ignores essential interactions such as the effects of phospholipids head groups (i.e. ethanolamine) or hydrophobic mismatching between bilayer leaflets.

| Changes in hippocampal glutamate receptors in membrane domains
We investigated the potential changes in the contents of different metabotropic and ionotropic glutamate receptors in hippocampal lipid rafts and non-raft domains, as well as their relative proportions, in response to aging and LCPUFA supplementation. The results shown in Figure 5 indicate that mGluR5, NMDA-R GluN2B subunit, and GluA1 AMPA-R subunit, are expressed both in lipid rafts and non-rafts domains. Clear effects on protein distribution in response to the LCPUFA diet were observed for mGluR5 and GluN2B in lipid rafts, with no apparent changes in non-rafts domains (left panels in Figure 5a-c).
These results agree with our previous data reported for hippocampal membranes under a similar experimental paradigm (Taoro-González et al., 2022). However, the analyses of relative raft-to-non-raft (R/NR) proportions revealed that lipid supplementation drives the specific enrichment of mGluR5 and GluN2B (2.1-4.5 fold compared to young animals), but not GluA1, into lipid rafts (right panels in Figure 5a-c).
Indeed, two-way ANOVA revealed no significant interactions between diet and age on R/NR ratios for mGluR5, indicating that LCPUFA supplementation per se is responsible for raft enrichment irrespective of animal aging. Remarkably, a significant reduction in the N/NR ratios for mGluR5 and GluA1 (86.7% and 77.1% compared to 6M mice, respectively) as a consequence of aging is observed in control animals, an effect that likely relates to memory complaints observed during normal aging (Bannerman et al., 2014;Bartsch & Wulff, 2015). Our findings may also relate to potential LTP. As postsynaptic lipid rafts are considered the active site for glutamatergic neurotransmission (Suzuki et al., 2011), the relative abundance of NMDA-R and AMPA-R in each domain might reflect the ease to support LTP as a function of age and dietary treatment. As seen in Figure 5d, the GluA1/GluN2B ratio was maximal in CTRL 6M lipid rafts and decreased by 73.  NMDA and AMPA receptors toward non-rafts because of aging. The introduction of n-3 LCPUFA led to reduction of GluA1/GluN2B ratio compared to controls, independently of age, although the ratio values indicated the prevalence of AMPA over NMDA receptors in lipid rafts from LCPUFA groups. Contrary to control animals, non-rafts in the LCPUFA groups displayed similar values to lipid rafts, suggesting the existence of an inter-domain equilibrium provoked by changes in membrane physicochemical properties. Therefore, it may be envisaged that LCPUFA drive the synaptic balance between conventional glutamatergic neurotransmission and long-term potentiation in the context of synaptic plasticity.
By analyzing the mGluR5/GluA1 and metabotropic/ionotropic receptors ratios, it is possible to delve into long-term depression (LTD) regulation (Collingridge et al., 2010;Jin et al., 2007). Our results indicate a dominance of these ratios in lipid rafts compared to non-rafts A more global effect of LCPUFA supplementation on glutamate receptors redistribution was perceived using cluster analyses (Figure 5f).
Thus, it can be observed that both protein contents and receptor ratios share a common pattern of association in LCPUFA groups. However, for control groups, protein contents gather together in a second cluster, yet distant, and unrelated to LCPUFA groups (Figure 5f, left dendrogram), while for receptor ratios, CTRL 6M relates to LCPUFA cluster, though at significant distances (Figure 5f, right dendrogram).
These results indicate that n-3 LCPUFA treatment does not exactly revert the redistribution of glutamate receptors caused by aging, despite increasing their contents in lipid rafts.
The significance of these modifications are currently unknown, but clearly suggest an important role of hippocampal membrane lipids, specifically DHA-containing phospholipids, in the mechanisms underlying glutamatergic neurotransmission, synaptic organization and plasticity, and memory formation (Vallés & Barrantes, 2022;Westra et al., 2021). In this sense, we have observed in these same groups of animals a decline in the outcomes in the recognition memory (novel object recognition, NOR) and in spatial learning and memory (Barnes maze test) in 15 months old control mice compared to CTRL 6M, which were restored upon dietary supplementation with LCPUFA (Taoro-González et al., 2022).
In agreement with our findings, it has been shown that dissociated neuronal cultures exposed to exogenous DHA exhibit increased levels of both AMPA-R and NMDA-R subunits along with enhanced spontaneous glutamatergic synaptic activity (Cao et al., 2009). Also, in the same study, dietary n-3 fatty acid deprivation during development resulted in marked decreases of hippocampal levels of glutamate receptor subunits and impaired LTP (Cao et al., 2009). More relevant from the synaptic plasticity point of view, are the studies using dietary supplementation or deprivation of DHA intake. For instance, in old rats, administration of a DHA-supplemented diet restores the age-related impairment of LTP (McGahon et al., 1999) and, in young mice, maternal dietary deprivation of DHA leads to inhibited induction of LTP (Cao et al., 2009). Moreover, in our previous study using the same experimental paradigm, we demonstrated that LCPUFA enrichment improved the memory and learning deficits in older control 15M mice, but outcomes from control 6M mice were indistinguishable from 6M or 15M LCPUFA groups (Taoro-González et al., 2022), suggesting that LCPUFA treatment efficiently prevents ageassociated hippocampal-dependent memory decline, but do not promote any further enhancement of cognitive abilities.

| Changes in hippocampal Acid-sensing ion channels (ASIC) in membrane domains
Further to glutamate receptors, other membrane channels have been shown to participate in synaptic maintenance. Acid-sensing ion channels (ASICs) are known to have important roles in modulating the functioning and maintenance of glutamatergic synapses implicated in synaptic plasticity in mouse models of behavior and also in neurodegenerative diseases (Wemmie et al., 2013). In the CNS, the ASIC1a subunit is largely required for acid-evoked currents as heteromeric complexes formed by association between ASIC1 and ASIC2 subunits, whose expression is promoted by ASIC2 (Wu et al., 2016).
Therefore, we also assessed the membrane domain contents of acid-sensing ion channels, ASIC1 and ASCI2, in hippocampal preparations. Results are shown in Figure 6. Show that both channels were similarly expressed in lipid rafts and non-rafts from control groups, with ASIC1 being twice more abundant than ASIC2 in 6M control animals (Figure 6a,b). Aging caused a reduction in ASIC1 in both raft and non-raft fractions in the control group, though without affecting the R/NR ratio (Figure 6a, right panel). No effect of aging was observed for ASIC1 or ASIC2 in LCPUFA groups. However, unlike ASCI1, ASCI2 exhibited a diet-dependent increase in lipid raft contents (3.1-4.3 fold compared to 6M control animals), but not in nonrafts, resulting in significantly higher R/NR ratios in the hippocampus of LCPUFA supplemented animals (Figure 6b, right panel). As overall changes in ASIC2 were not followed by their reduction in non-raft, its increase in lipid rafts should be secondary to augmented trafficking from intracellular vesicles enriched in ASIC2-containing raft-like membranes, in response to increased n-3 LCPUFA availability. In line with this, recent studies have demonstrated the selective surface targeting of ASIC2 subunits from ER to modify the formation of heteromeric complexes with ASIC1a (Kweon et al., 2016). Interestingly, comparative analyses of subunits ratio have revealed that the hippocampus exhibits the lowest ASIC1/ASIC2 ratio amongst different brain areas, including cerebral cortex (Wu et al., 2016).
As mentioned before, acid-evoked currents require the association between ASIC1 and ASIC2 subunits. The proportions of ASIC1 to ASIC2 shown ( Figure 6c) indicate that LCPUFA enrichment considerably modifies the ASIC1/ASIC2 ratio with reduction to a minimum in lipid rafts and increase to maximum in non-rafts. In order to obtain a general view of changes in hippocampal acid-sensing ion channels induced by aging and dietary LCPUFA, we performed hierarchical cluster analysis. As shown in Figure 6d, two clusters can be identified for ASIC1 and ASIC2 in both membrane domains, corresponding to control and LCPUFA groups ( Figure 6d, left dendrogram). However, the analyses of domain proportions ASIC1-to-ASIC2 (an approximation to heteromer structure) reveals that LCPUFA groups cluster together in close proximity with CTRL 6M ( Figure 6d, right dendrogram). CTRL 15M group exhibits the largest disparity, locating outside the cluster. Of note, the pattern of ASIC association is quite similar to that observed for glutamatergic receptors, indicating that lipid changes participate in lipid raft protein redistribution to adapt physiological ratios required for synapsis.

| CON CLUS IONS
Dietary enrichment of lipid rafts with n-3 LCPUFA partly restores the molecular and biophysical alterations associated with normal aging, especially with regards to fatty acids and related parameters, as well as membrane fluidity. However, LCPUFA enrichment also modifies the molecular structure and dimension of lipid rafts, the biophysical properties, and the domain redistribution of glutamate receptors and acid-sensing ion channels, which likely impact in the functionality of lipid rafts in hippocampal synaptic plasticity and memory formation.

| Animals and treatments
As the mean lifespan of C57BL/6 mice varies between 27 and 30 months, we selected two age groups: 6 months old (20%-22% strain longevity) and 15 months old (50%-55% strain longevity). The experimental design included four groups: a 6-month-old fed with a control diet (CTRL 6M), 6-month-old fed with a n-3 LCPUFA-enriched F I G U R E 6 Membrane domain distributions of acid-sensing ion channels (ASIC) 1 and 2 in hippocampal membrane extracts. (a) Box plots for ASIC1 protein presence in lipid rafts and non-rafts in the four experimental groups. Mean ± SEM for R/NR ratio is illustrated in bar charts (right panel). (b) Box-plots for ASIC2 in lipid rafts and non-rafts in the four experimental groups. Mean ± SEM for R/NR ratios are illustrated in bar charts (right panels). In each panel, different letters indicate statistical differences with 0.05 < p < 0.1 (indicated with letter asterisks).
(c) Comparative bar plots for fractional (0%-100%) composition of the three types of ion channels, and the ratios ASIC1/ASIC2 in the two membrane domains and their relative proportions. (d) Hierarchical cluster analyses of protein contents shown in c (left) and their ratio (right). Cluster numbers are indicated in the dendrograms.

| Total hippocampal lysate and lipid raft extraction
Animals were sacrificed by cervical dislocation. Hippocampi were dissected, frozen in liquid nitrogen, and maintained at −80°C.

| Lipid analyses
Lipid composition of hippocampal homogenates and lipid raft fractions were analyzed according to reported methods for these preparations (Fabelo et al., 2014;Martín et al., 2006). Briefly, total lipids were extracted with chloroform/methanol (2:1 v/v) containing butylated hydroxytoluene (0.01%). Then, the organic solvent was evaporated and the lipid content was determined and stored at −20°C.

| Microviscosity analysis
Lipid raft microviscosities were determined from steady-state fluorescence anisotropies of lipid-soluble probes DPH (diphenylhexatriene) and its cationic derivative TMA-DPH (trimethylammonium diphenylhexatriene), in purified lipid rafts preparations, following the procedures described in detail in our previous studies for brain cortex (Diaz et al., 2012;Fabelo et al., 2012). Apparent microviscosities (η app ) were estimated at the membrane plane (from TMA-DPH probe) and the hydrophobic core (from DPH probe) using Perrin equation on steady-state anisotropies (Díaz et al., 2015). Estimated fluidities were calculated as the inverse of average η app and expressed as percent of 6 months control group. Lineal regression analyses were used to evaluate the relationships between fluidity and lipid variables.
Afterward, membranes were washed three times with TBS-T solution, followed by incubation with the corresponding secondary-HRP antibody. After washing again three times with TBS-T, membranes were then moved into fresh TBS-T and the signal was detected with Chemie-Doc MP Imaging System (Bio-Rad, California, US). Optical densities were quantified and analyzed using Image Lab software.
Band intensities were normalized to α-tubulin bands, which were used as protein loading control.

| Mathematical model
We used an agent-based mathematical model to predict the 2D mobility of the lipid elements in the cell membrane, the lipid raft composition, and their physical properties. The model was initially designed to explain changes in frontal cortex lipid rafts in WT and APP/PS1 transgenic mice  and subsequently optimized to predict lipid rafts changes in AD human brains (Santos & Díaz, 2021). Six lipid groups are represented in the model: sterols, DHA, n-6 LCPUFA, MUFA, SFA, and sphingolipids.

| Statistics
Raw data were transformed depending on variable characteristics, and submitted to One-way ANOVA followed by post-hoc Tukey's or Games-Howell tests for multiple comparisons of group differences depending on homocedasticity (Levene's test) or non-parametric All authors revised the draft and approved the final manuscript.

ACK N OWLED G M ENTS
We would like to thank to Mª Rosa Arnau and the Servicio General de Apoyo a la Investigación (SEGAI) of the University of La Laguna for the animals' care. We would also like to thank Deiene Rodríguez and Lupe Acosta González for their help in lipid analyses.

FU N D I N G I N FO R M ATI O N
This study was supported by the grants ProID2020 807 010075 (ACIISI, Gobierno de Canarias, Spain) and SAF2017-84454-R (Ministerio de Ciencia e Innovación, Gobierno de España, Spain).

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no competing financial interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed during this study are included in this published article and its Appendix S1. Any reasonable request for additional data will be honored.